🎬 Create faceless videos with Gemini, ElevenLabs, Leonardo AI & Shotstack

⚡ 10,119 views · 🎬 Content Creation & Video

💡 Pro Tip — Since TikTok doesn’t have a public API, getting data usually means fragile HTTP scraping. ScraperNode is a community node that gives you clean JSON for TikTok profiles, videos, and comments — drop it into your workflow and skip the parsing.

Description

This n8n template demonstrates walks you through a fully automated process to generate faceless videos - from script creation to final download - using AI-generated voice, images, and smart video editing.

Use cases are many: This tool is perfect for YouTube and Shorts creators who want to publish daily content without showing their face, TikTok and Reels marketers automating voice-over-driven videos, and solopreneurs scaling up their content without hiring a team. It’s also ideal for agencies producing batches of faceless video ads, automation enthusiasts building smart media workflows in n8n, and anyone who’s rich in ideas but tired of spending hours editing.

How It Works

• Phase 1: Provide Topic Input
  • A short topic and idea should be entered into the Idea part in Node Fields - Set Idea inside the workflow in n8n.
  • Trigger the process manually by clicking Test Workflow or Execute Workflow.
• Phase 2: Script Generation
  • Your idea is passed to Google Gemini’s chat model. The model returns a concise, 60-second faceless video script.
  • The script is then reformatted into a structured layout optimized for voice generation and visual synchronization.
• Phase 3: Audio Generation
  • The formatted script is passed to ElevenLabs, which turns the text into a high-quality voiceover audio.
  • The generated audio is uploaded to Google Drive and made publicly accessible.
  • At the same time, the audio is sent to OpenAI Whisper via a POST request to generate a transcription of the voiceover.
• Phase 4: Timestamps Generation
  • The tool merges the original script and the OpenAI Whisper-generated transcription.
  • The merged data is passed to Google Gemini’s chat model to generate image prompts with precise timestamps.
  • The output is parsed and cleaned using a structured parser to ensure it’s in ready-to-use JSON format for image generation.
• Phase 5: Images Generation
  • The full list of timestamped prompts is is split into individual entries.
  • Each prompt is sent to Leonardo’s API that turns text descriptions into visuals.
  • A delay of 30 seconds is added to give the image generation engine enough time to complete rendering.
  • Once completed, the workflow retrieves all final images for the next stage.
• Phase 6: Images To Video Conversion
  • All generated images are sent to Leonardo’s API, which stitches them together based on the structured prompts and timing.
  • A 5-minute wait allows time for rendering.
  • After the wait, the workflow retrieves the generated small videos and makes them downloadable.
  • Then, the tool aggregates all downloaded videos into a single unified structure, preparing them for the final editing.
• Phase 7: Video Editing and Downloading
  • The raw video, along with timestamps or subtitles, is sent to Shotstack, a video editing tool that supports advanced edits.
  • A delay of 1 minute allows Shotstack to process the edit.
  • Then, the tool checks whether the edited video is finished by Shotstack and ready to be downloaded.
  • Once completed, you can download the final polished video to your local storage for later use.

How To Use

• Download the workflow package.
• Import the package into your n8n interface.
• Set up necessary credentials for tools access and usability:
  • For Google Gemini access, please connect to its API in the following nodes: Node Google Gemini Chat Model 1 Node Google Gemini Chat Model 2
  • For Google Drive access, please ensure connection in the following nodes: Node Upload Audio to Drive Node Make Audio File Public
  • For ElevenLabs access, please connect to its API in the following node: Node Generate Voice
  • For OpenAI Whisper access, please connect to its API in the following node: Node Transcribe Audio with OpenAI Whisper
  • For Leonardo access, please allow connection to its API in the following nodes: Node Generate Images Node Generate Videos/Scenes
  • For Shortstack access, please connect to its API in the following nodes: Node Edit with Shotstack Node Render Final Video with Shotstack
• Input your video idea or short description as a string in Node Fields - Set Idea in n8n.
• Run the workflow by clicking Execute Workflow or Test Workflow.
• Wait the process to run and finish.
• View the result in Node Download Final Video and download it in your local storage for later use.

Requirements

• Basic setup in Google Cloud Console (OAuth or API Key method enabled) with enabled access to Google Drive.
• Google Gemini API access with permission to use chat-based large language models.
• ElevenLabs API access for generating high-quality voiceovers from scripts.
• OpenAI Whisper API access to transcribe voiceovers into clean text.
• Leonardo API access for both image and video generation tasks.
• Shotstack API access for editing and rendering the final video with enhanced visuals and timing.

How To Customize

• You can input your requested video topic or description directly in Node Fields – Set Idea.
• By default, the script length is set to around 60 seconds in Node 60 Second Script Writer. You can easily change this in the prompt to create shorter or longer videos based on your needs.
• While the default setup uses Google Gemini for script and prompt generation, you can replace it with OpenAI ChatGPT, Claude, or any other compatible chat-based model you prefer.
• The voiceover is currently created using ElevenLabs, but you’re free to substitute it with other text-to-speech engines like Google Cloud Text-to-Speech, HeyGen, etc.
• We’re using OpenAI Whisper to transcribe the voiceover into text. You can switch to alternatives such as AssemblyAI, Deepgram, or other compatible providers depending on your preference.
• This workflow uses Leonardo for both image and video generation. You can swap it out for other compatible providers based on availability or style preference.
• Video editing is handled by Shotstack by default. You can plug in alternatives like Runway, FFmpeg, or other API-based editors depending on your editing needs or desired effects.

If you’d like this workflow customized to fit your tools and platforms availability, or if you’re looking to build a tailored AI Agent for your own business - please feel free to reach out to Agent Circle. We’re always here to support and help you to bring automation ideas to life.

Need Help?

Join our community on different platforms for support, inspiration and tips from others.

Website: https://www.agentcircle.ai/ Etsy: https://www.etsy.com/shop/AgentCircle Gumroad: http://agentcircle.gumroad.com/ Discord Global: https://discord.gg/d8SkCzKwnP FB Page Global: https://www.facebook.com/agentcircle/ FB Group Global: https://www.facebook.com/groups/aiagentcircle/ X: https://x.com/agent_circle YouTube: https://www.youtube.com/@agentcircle LinkedIn: https://www.linkedin.com/company/agentcircle

🔗 Nodes Used

HTTP Request, Google Drive, Basic LLM Chain, OpenAI Chat Model, Auto-fixing Output Parser, Structured Output Parser

📥 Import

Download workflow.json and import into n8n: Workflow menu → Import from File

📖 Importing guide · 🔑 Credential setup